Mar 12, 1996 - Jefferson Cancer Institute, Thomas Jefferson University, Philadelphia,. PA 19107 .... (PI3-kinase)-like domain (Hartley et al., 1995), and it is.
The EMBO Journal vol.15 no.13 pp.3238-3246, 1996
DNA-dependent protein kinase catalytic subunit: a target for an ICE-like protease in apoptosis
Qizhong Song1, Susan P.Lees-Miller2, Sharad Kumar3, Ning Zhang1, Doug W.Chan2, Graeme C.M.Smith4, Stephen P.Jackson4, Emad S.Alnemri5, Gerald Litwack5, Kum Kum Khanna1 and Martin F.Lavin1"6'7 'Queensland Cancer Fund Research Unit, Queensland Institute of Medical Research, Bancroft Centre and 6Department of Surgery, University of Queensland, PO Royal Brisbane Hospital, Herston. 4029 Brisbane, Australia, -Department of Biological Sciences, University of Calgary, Calgary, Alberta, T2N IN4, Canada, 3Hanson Centre for Cancer Research, Box 14, Rundle Mall Post Office, Adelaide, SA 5000, Australia, 4Wellcome/CRC Institute and Department of Zoology, University of Cambridge, Tennis Court Road, Cambridge CB2 IQR, UK and 5Department of Pharmacology and the Jefferson Cancer Institute, Thomas Jefferson University, Philadelphia, PA 19107, USA 7Corresponding author
Radiosensitive cell lines derived from X-ray cross complementing group 5 (XRCC5), SCID mice and a human glioma cell line lack components of the DNAdependent protein kinase, DNA-PK, suggesting that DNA-PK plays an important role in DNA doublestrand break repair. Another enzyme implicated in DNA repair, poly(ADP-ribose) polymerase, is cleaved and inactivated during apoptosis, suggesting that some DNA repair proteins may be selectively targeted for destruction during apoptosis. Here we demonstrate that DNA-PKcs, the catalytic subunit of DNA-PK, is preferentially degraded after the exposure of different cell types to a variety of agents known to cause apoptosis. However, Ku, the DNA-binding component of the enzyme, remains intact. Degradation of DNAPKcs was accompanied by loss of DNA-PK activity. One cell line resistant to etoposide-induced apoptosis failed to show degradation of DNA-PKcs. Protease inhibitor data implicated an ICE-like protease in the cleavage of DNA-PKcs, and it was subsequently shown that the cysteine protease CPP32, but not Mch2a, ICE or TX, cleaved purified DNA-PKcs into three fragments of comparable size with those observed in cells undergoing apoptosis. Cleavage sites in DNA-PKcs, determined by antibody mapping and microsequencing, were shown to be the same for CPP32 cleavage and for cleavage catalyzed by extracts from cells undergoing apoptosis. These observations suggest that DNA-PKcs is a critical target for proteolysis by an ICE-like protease during apoptosis. Keywords: apoptosis/DNA-dependent protein kinase/ etoposide/ICE-like protease/substrate
Introduction Apoptosis is an ultrastructurally distinct form of cell death that occurs in response to a variety of stimuli (Kerr et al., 3238
1972; Wyllie et al., 1980). Cell death by apoptosis is essential for the maintenance of homeostasis during development and in adult tissues (reviewed in Vaux, 1993; Vaux et al., 1994; Steller, 1995). Cells undergoing apoptosis are characterized by a contraction of the cytoplasm, plasma membrane blebbing, nuclear condensation and fragmentation of DNA (Kerr et al., 1972; Wyllie et al., 1980). The regulation of this process is complex and involves the expression of a number of genes conserved throughout the phylogenic scale capable of inhibiting or activating apoptosis (Strasser, 1995; White, 1996). The discovery of several genes responsible for regulatory cell death in Caenorhabditis elegans has provided major insights into the mechanisms involved. These genes, which include ced-3, ced-4 and the bcl-2-like gene ced-9, control various elements of the programmed cell death pathway in C.elegans (Ellis and Horvitz, 1986; Hengartner et al., 1992). The ced-3 gene is essential for developmentally programmed cell death in C.elegans (reviewed in Ellis et al., 1991), and the protein it encodes was shown to share homology with mammalian interleukin- 1 -converting enzyme (Yuan et al., 1993), a novel cysteine protease required for the processing of pro-ILI18 (Thornberry et al., 1994). Subsequently, several other mammalian proteins similar to ICE/CED-3 have been identified (Fernandes-Alnemri et al., 1994; Kumar et al., 1994, 1995; Lazebnik et al., 1994; Wang et al., 1994; Faucheu et al., 1995; Fernandes-Alnemri et al., 1995a,b; Munday et al., 1995; Nicholson et al., 1995; Tewari et al., 1995a), and now there is substantial evidence that these proteases play an essential role in apoptosis (reviewed in Kumar and Harvey, 1995; Kumar et al., 1995; Martin and Green, 1995; Kumar and Lavin, 1996). Poly (ADP-ribose) polymerase (PARP) is cleaved into an 85 kDa polypeptide in cells undergoing apoptosis in response to a variety of treatments. It is now known that this occurs by the action of the ICE-like proteases prICE/CPP32/Yama/apopain and Mch3 (Lazebnik et al., 1994; Fernandes-Alnemri et al., 1995b; Nicholson et al., 1995; Tewari et al., 1995a). PARP cleavage can also be mediated by Nedd2, ICE, TX (Gu et al., 1995) and Mch2 (Fernandes-Alnemri et al., 1995a), albeit with lower efficiency. Other proteins thought to be targets for ICE-like proteins in apoptotic cells include nuclear lamins (Lazebnik et al., 1995), actin (Mashima et al., 1995), U1 -70 kDa (Casciola-Rosen et al., 1994; Tewari et al., 1995b), Gas 2 (Brancolini et al., 1995) and PKC6 (Emoto et al., 1995). Clearly, apoptotic events require the coordinate degradation of certain crucial target proteins, and to understand the mechanism of apoptosis, it is crucial to identify death substrates and the responsible proteases. We have attempted to identify other potential key substrates for apoptotic proteolysis based on their known Oxford University Press
DNA-dependent protein kinase and apoptosis
a
b
BL30A + etoposide (68 ;,.NI)
0
lime th)
2
1
3
4
5
li'iethilli
(1
X1
2
:3
4
I
nftA1ADK,r
(460 kD)a 240 kDa
_ _ _ _-
-
flragmoient
C Time (h)
Ku80 _.Ku70 _
0
_
1
2
3
4
-__ -
-
_A
-
5
-
Fig. 1. Cleavage of DNA-PKcs occurs during apoptosis induced by etoposide in Burkitt's lymphoma cells. DNA-PKcs was detected using polyclonal antibodies prepared against a peptide containing amino acids 2018-2136 (DPKI). (a) Time course of cleavage of DNA-PKcs after the exposure of BL30A cells to 68 tM etoposide. (b) Time course of DNA fragmentation after the treatment of BL30A cells with 68 lM etoposide. (c) Lack of degradation of Ku7O and Ku8O, the DNA-binding components of DNA-PK, after the treatment of cells with etoposide. Ku was detected by immunoblotting with a mouse polyclonal antibody.
functions. One such protein is DNA-dependent protein kinase (DNA-PK), which is involved in double-strand break repair and V(D)J recombination. An enzyme such as this, which is involved in DNA repair, might be expected to be preferentially degraded during DNA fragmentation that accompanies apoptosis. DNA-PK comprises a catalytic subunit of 460 kDa, called DNA-PKcs, and a DNA-binding component termed Ku, a heterodimer of 70 and 80 kDa subunits (Dvir et al., 1992; Gottlieb and Jackson, 1993). It has been demonstrated recently that DNA-PKcs contains a phosphatidylinositol 3-kinase (PI3-kinase)-like domain (Hartley et al., 1995), and it is related to a number of other proteins involved in DNA damage response and cell cycle control (Hari et al., 1995; Jackson, 1995; Lavin et al., 1995). A member of this family is the protein mutated in ataxia-telangiectasia (ATM; Savitsky et al., 1995), a syndrome characterized by radiosensitivity and cancer predisposition (Sedgwick and Boder, 1991). Here we demonstrate that DNA-PKcs is degraded into specific fragments during the process of apoptosis in several different cell types undergoing apoptosis, and in response to different apoptotic agents. We show that degradation causes loss of activity of the enzyme, and that an ICE-like protease is involved in the proteolytic process.
Results Degradation of DNA-PKcs in etoposide-treated BL30A cells To determine whether DNA-PKcs might be a target for proteolytic degradation during DNA damage-induced apoptosis, we used immunoblotting with DPK1, an antibody raised against amino acid region 2018-2136 of the protein (Lees-Miller et al., 1995), with extracts from a Burkitt's lymphoma cell line, BL30A, previously shown to be susceptible to apoptosis (Filippovich et al., 1994). The treatment of BL30A cells with etoposide (68 ,uM), followed by immunoblotting with DPKl, revealed that
30-40% of DNA-PKcs was degraded to an intermediatesized form, ~240 kDa, by 3 h post-treatment (Figure la), and by 5 h ~80-90% of the protein was degraded. Under these conditions, proteolysis appeared to be highly selective because high molecular weight proteins (>200 kDa) viewed by Coomassie blue staining of the gel remained intact (results not shown). The degradation of DNA-PKcs paralleled DNA fragmentation into oligonucleosomal sized pieces (Figure lb), and morphological changes indicative of apoptosis revealed that 26% of cells were undergoing apoptosis at 3 h post-treatment and that this had risen to 71% by 5 h. Since DNA-PKcs must associate with Ku, the DNAtargeting component of the complex, to be activated (Gottlieb and Jackson, 1993), we also determined the fate of Ku7O and Ku8O during apoptosis. There was no evidence for the degradation of either protein in cells exposed to 68 jM etoposide for up to 5 h post-treatment (Figure lc).
DNA-PKcs cleavage is a common feature of apoptosis The exposure of BL30A cells to a variety of agents shown previously to induce apoptosis in these cells resulted in the degradation of DNA-PKcs to the same intermediatesized fragment (Figure 2a). In addition, this cleavage was observed in three other Burkitt's lymphoma cell lines exposed to etoposide (Figure 2b). On the other hand, no degradation was observed in BL29, a cell line resistant to etoposide-induced apoptosis (Figure 2b; lanes 7 and 8). This specific cleavage of DNA-PKcs was also demonstrated in HeLa, U937 and Molt-4 cells exposed to different agents (Figure 2c). The effect of proteolysis on DNA-PK activity To determine whether the cleavage of DNA-PKcs interferes with its function, we measured the activity of DNAPK in BL30A cells treated with etoposide using a peptide substrate (Lees-Miller et al., 1995). It is evident from the 3239
- - - - - - lr- Xr
Q.Song et al
C .
a
a
E. to poS i dI
.E~
.9
I-
d
-r X a* 1i;t i X+ 1
.-
ic0 ki1i)~119 l)NAlkl w0_
0_
-
2 1U) kI),, frrlimerIt
-
em
240 kDaImmfragment
Etoposide D)NA-PKcs (460 kDa) _ 240 kiDas fratgnejit
BMI3674
OK 11--+
-
d_
+
-
B1,29
all cases, for 12 h.ofInDNA(10 ,ug/ml) h; or underwent (v) tetrandrine 12 cells incubated for of the (b) Cleavage -70-80% apoptosis. PKcs in the presence (+) or absence (-) of 68 ,uM etoposide in different Burkitt's lymphoma cell lines. (c) Cleavage of DNA-PKcs in HeLa, U937 and Molt-4 cells undergoing apoptosis induced by a_4dIP ~different agents. For HeLa cells, extract was prepared from detached +
-
+
results in Figure 3 that kinase activity is reduced to 60% upon treating cells with etoposide for 3 h. At this timepoint, a significant amount of proteolytic degradation had taken place (see Figure la). Furthermore, after 5 h, activity was reduced to ~25% that of untreated controls, and it continued to decline with time; by 24 h no DNA-PK activity was detectable. The loss of DNA-PK activity therefore correlates with the degradation of DNA-PKcs.
cells after 6 h of treatment with
Identification of the protease that cleaves DNA-PKcs We initially employed a series of inhibitors to identify the class of protease involved in the cleavage of DNA-PKcs. The degradation of DNA-PKcs in BL30A cells treated with etoposide was not prevented by the protease inhibitors leupeptin, aprotinin, pepstatin, phenylmethylsulfonyl fluoride (PMSF) and E-64 (Figure 4a). However, tosyl-Lphenylalanine chloromethyl ketone (TPCK) and tosyl-Llysine chloromethyl ketone (TLCK) completely prevented apoptosis and the degradation of DNA-PKcs in BL30A cells (Figure 4a). Other inhibitors are iodoacetamide and N-ethylmaleimide. Because these are toxic in BL30A cells, extracts were prepared from etoposide-treated cells, and inhibitors and purified DNA-PKcs (as substrate) were added. Under these conditions, iodoacetamide, Nethylmaleimide and YVAD-chloromethylketone (CMK), a specific inhibitor of ICE-like proteases, completely prevented the degradation of DNA-PKcs (Figure 4b). This pattern of inhibition suggested that an ICE-like protease is involved in the degradation of DNA-PKcs. A Northern blot analysis (Figure 5) demonstrated that ICE, CPP32, Mch2a, TX and NEDD2 were all expressed in this cell line. Exposure to a radiation dose (10 Gy) that causes apoptosis in BL30A did not lead to any significant changes in mRNA expression for the various proteases, suggesting
etoposide (68
,uM).
Extract
was
also
prepared at the same time from etoposide-treated U937 cells; in the case of Molt-4, cells were irradiated (20 Gy) and extracts prepared at 12 h post-irradiation. 1 20
-
i 00
K
~~~X.aX SH////SxtlW///~~~~~ 1' 80I Vt4V,4 ,4/ .'f
C.)
a._
3240
..
Fig. 2. DNA-PKcs cleavage in different cell types by a variety of agents causing apoptosis. (a) Effect of different agents that cause apoptosis on the cleavage of DNA-PKcs in BL30A cells. Cells were treated with: (i) 5% ethanol for 12 h; (ii) EGTA (5 mM) for 12 h; (iii) cisplatin (20 ,ug/ml) for 24 h; (iv) y-irradiation (20 Gy) and
B129 z'
BM13674
WW2
BL3OK
BI
mm_
_
_
el
b (:ell line
+
+
E-
4
(460 kDa)
I_>
i
Co
5~~~~~~~~~u.
l)NA-PKcs
Nlolt-t
U937
If.Ill
(eli lineI'
CL
60 *SUJVSxJ:~~zii
IV/AE /AWAVKIL ~~~~~~~JVWAW/A~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~..~%V/1K¾44 4
z 0
2 0/ / 0g
>
0
1
2
3
4
5
24
TIME (h)
Fig. 3. DNA-PK activity in etoposide-treated BL30A cells undergoing apoptosis. For DNA-PK activity, extracts were prepared and assayed with a synthetic peptide (PESQEAFADLWKK), as described previously (Allalunis-Turner et al., 1995).
that the activation of an existing protease was responsible for the degradation of DNA-PKcs. To investigate which protease might be responsible, we exposed purified DNAPKcs to several proteases (isolated as bacterial lysates, as described in Materials and methods) to identify a potential candidate for cleavage of this protein. Interestingly, CPP32 cleaved purified DNA-PKcs (Figure 6, lane 3) to produce a fragment of comparable size with that observed in cells undergoing apoptosis (lane 2). In contrast, neither Mch2a (lane 4), TX (lane 5) nor ICE (lane 6) were capable of cleaving DNA-PKcs in this way.
DNA-dependent protein kinase and apoptosis
co
a
4)
.Z
...I.0
*4)0 a m~~~~~~~~~~~~~~~
M
=
b m
DNA-PKcs (460 kDa) 240 kDa
_ _
PC
DNA-PKces -.
a-N
-
fragment
(460 kDa)
NO
240 kDa fragment
a-
6W
am
_
Fig. 4. Effect of protease inhibitors on DNA-PKcs cleavage in BL30A cells undergoing apoptosis. (a) Cells were exposed to etoposide alone or to etoposide in the presence of the inhibitors PMSF (570 gM), leupeptin (210 tM), pepstatin (3.6 tM), aprotinin (0.2 ,uM), E-64 (50 ,uM), TPCK (200 tM) and TLCK (200 gM). Pre-incubation of BL30A cells with etoposide (68 gM) was carried out for 30 min. These cells were washed prior to exposure to inhibitors, followed by harvesting 4.5 h later. Extracts were prepared for immunoblotting and DNA-PKcs detected with DPKI antibody. (b) Effect of inhibitors on the degradation of purified DNA-PKcs by extracts prepared from cells undergoing apoptosis. N-Ethylmaleimide (5 mM), YVAD-CMK (10 giM) and iodoacetamide (5 mM) were used with cell extracts prepared from cells exposed to etoposide. Lane 1, extracts prepared from untreated BL30A cells added to DNA-PKcs; lane 2, extracts prepared from BL30A cells undergoing apoptosis added to DNA-PKcs; lanes 3-5, extracts from etoposide-treated cells preincubated with the different inhibitors pfior to incubation with DNA-PKcs.
Antibody mapping and microsequencing A series of antibodies directed against different regions of DNA-PKcs were employed to study the overall pattern of degradation of the protein in cells undergoing apoptosis and to localize the approximate break points (Figure 7a). Polyclonal antisera (AbFLA) against the whole DNAPKcs molecule detected a fragment of -240 kDa in addition to a fragment of 150 kDa and a weakly crossreacting band of 120 kDa (Figure 7b). A polyclonal antibody (Ab9607) against region 2790-3065 crossreacted strongly with the 150 kDa fragment and very weakly with the 240 kDa fragment, which may be a nonspecific cross-reaction. It was not possible to detect the 120 kDa band with this antibody. A monoclonal antibody (Ab42-27; Carter et al., 1990) against the kinase domain detected only the 150 kDa fragment and a much weaker 120 kDa fragment. A weak band of lower molecular weight was also evident, but this appeared to be nonspecific because it was also present in the untreated sample. Finally, an antibody directed against amino acid residues 1550-1840 (AbSLY) reacted only with the 240 kDa fragment. This information, together with the data obtained with the DPK1 antibody, indicate that the 150 kDa fragment is towards the C-terminus of the molecule and the 240 kDa segment is internal in the protein and contiguous with the 150 kDa fragment. Based on the antibody mapping, it is likely that one cleavage site occurs ~150 kDa from the C-terminus. A candidate cleavage site for an ICE-like protease exists at position 2709-2713 (DEVD4kN). The yield of the 120 kDa fragment was normally quite low and appears to arise from the 150 kDa fragment with time because it is recognized by AbFLA (against the whole protein) and Ab42-27 which detects the kinase domain at the C-terminus. Failure to detect a fragment(s) from the N-terminal region of DNAPKcs could mean that it is degraded rapidly into low molecular weight fragments. Because CPP32 also generated a 240 kDa fragment from purified DNA-PKcs, we expected that the other fragments would be formed if this enzyme was responsible for the degradation of DNA-PKcs in vivo. After the digestion of DNA-PKcs with CPP32, polyclonal AbFLA
BI.30A) + v-irradiationi (IO(J-) Timle (h)
('PP32
M"CH2;
I*+
N-f~..D. ~ ~ F
ICE
GA PD!IH
*:
*
Fig. 5. Expression of mRNAs for ICE-like proteases in BL30A cells treated with y-radiation (10 Gy), as determined by a Northem blot analysis. Poly(A)+ RNA was prepared using a Pharmacia QuickPrep Micro mRNA kit and separated by electrophoresis prior to transfer onto Hybond N membranes (Amersham). Blots, prepared under similar conditions, were hybridized with the various probes. CPP32, TX and Mch2 probes detected 2.6, 1.5 and 1.4 kb transcripts, respectively. The NEDD2 probe detected a 4.2 kb message and a slightly larger transcript. The ICE probe hybridized to transcripts of 1.7 and 0.5 kb in size. Only the prominent species (1.7 kb) is shown. Under the hybridization conditions used, no signals were detected for ICE,eIII.
recognized fragments of 240 and 150 kDa and also gave a strong signal for a 120 kDa fragment (Figure 8, lane 3). Comparison with the digestion pattern of DNA-PKcs in extracts from cells undergoing apoptosis revealed a concordance in the 240 and 150 kDa bands (Figure 8, lanes 2 and 3). It was not possible to detect the 120 kDa band in the extract lane in this experiment. The difference in the cross-reactivity of the 120 kDa fragment may be caused by additional digestion of this fragment in vivo 3241
Q.Song et al.
compared with in vitro; as seen previously (Figure 7b), less of the 120 kDa is generated under in vivo than in vitro conditions. This pattern of digestion was observed only when DNA and Ku, the DNA binding component of DNA-PK, were present in the reaction mix, suggesting that during apoptosis, the cleavage of DNA-PKcs occurs when it is associated with DNA. Furthermore, under these conditions Ku was not degraded by CPP32, as was also the case in vivo (data not shown). An N-terminal sequence analysis of the 150 kDa fragment generated by CPP32 digestion of DNA-PKcs identified the cleavage site between Asp2712 and Asn2713, which corresponds to the DEVD4IN site predicted from the antibody mapping data. Cleavage at this site is also observed when DNA-PKcs
-.
.4Fow
z
+ 0
0
C
DNA-PKcs (460 kDa) 240 kDa fragment
-
-
Fig. 6. Cleavage of purified DNA-PKcs by ICE and ICE-like proteases. DNA-PKcs was incubated with purified ICE or bacterial cell extracts containing CPP32, Mch2a or TX for 30 min at 37°C, as described in Materials and methods. DNA-PKcs cleavage was detected by a Western blot analysis.
is incubated with extracts prepared from BL30A cells undergoing apoptosis. A sequence analysis of the 120 kDa fragment identified a cleavage site between Asp2982 and Gly2983 in the sequence DWVD4.G (2979-2983), which is within the 150 kDa fragment, and would be expected to generate an additional fragment of 120 kDa similar to that observed when the 150 kDa fragment is cleaved
further.
Discussion Recent studies have demonstrated that the ICE family of proteases are important components in the execution phase of apoptosis (Fernandes-Alnemri et al., 1995a,b; Kumar, 1995a; Martin and Green, 1995; Munday et al., 1995; Nicholson et al., 1995). It is evident that the activation of these proteases in turn causes the cleavage of a number of key protein targets essential for apoptosis. Although several ICE-like proteases have been identified and a few protein targets have been reported to be degraded during apoptosis, the interaction between these remains largely undefined. The only protein shown to be an in vitro target for an identified ICE-like protease is PARP (Lazebnik et al., 1994; Fernandes-Alnemri et al., 1995a,b; Nicholson et al., 1995; Tewari et al., 1995a). The identification of new substrates and the enzymes responsible for their cleavage, together with information on the reactivity and specificity of these enzymes, will provide an insight into the mechanism of apoptosis. We have demonstrated here that DNA-PKcs is specifically degraded in a variety of different cell types undergoing apoptosis in response to different cytotoxic agents. Importantly, one cell line resistant to apoptosis failed to show any evidence of the degradation of DNA-PKcs after exposure to etoposide. Cleavage of DNA-PKcs in apoptotic cells was accompanied by a loss of enzyme activity, which
a AbSLY DPKI N-term inus
Ab96O7
Ab42-27 C-terminus
1.
C00o
2000
3C00
4000
AbFLA
b
Ab FLA
460 kDa
_
240 kDa
_
1l0 kD
_-
120 kDa
_
Ab 967
Ab 42-27
Ab SLY 46) k D
240 kD
24.0 kDa
-
I1150 kDa I.-
12) kDu 120 kDA
-
_
m-_ _0
Fig. 7. Localization of the sites of cleavage in DNA-PKcs in cells undergoing apoptosis using a panel of antibodies specific for different regions of the protein. (a) Recognition sequences for the different antibodies. DPKI recognizes amino acids 2016-2136; Ab9607 recognizes amino acids 27903065; AbSLY recognizes amino acids 1550-1840; monoclonal Ab42-27 recognizes the kinase domain; AbFLA is an antibody against the whole molecule. (b) Immunoblotting with antibodies specific for different regions of the molecule. Extracts were prepared as described in Materials and methods from BL30A cells undergoing apoptosis, separated by 8% SDS-polyacrylamide gels and immunoblotted with the appropriate antibody. The left lane in each case represents untreated and the second lane extracts from etoposide-treated cells.
3242
DNA-dependent protein kinase and apoptosis
0
+ Po
464) kDa 240 kDa
_ _
150 kDa 120 kDa
_ _
Po
eQ
Pi
Fig. 8. Cleavage pattern of purified DNA-PKcs by CPP32 compared with the degradation of DNA-PKcs in etoposide-treated cells immunoblotted with AbFLA. Purified DNA-PKcs was incubated with 2 pg CPP32 bacterial lysate as described in Materials and methods, and extracts were prepared from cells undergoing apoptosis prior to separation on 8% SDS-polyacrylamide gels.
can be explained by cleavage of the only known functional domain, the kinase catalytic domain, from the rest of the molecule (Hartley et al., 1995). This cleavage may also disrupt other as yet undescribed functions of DNA-PK. The use of iodoacetamide, N-ethylmaleimide and TyrVal-Ala-Asp-chloromethylketone (YVAD-CMK), as well as the results in cell extracts, suggest that an ICE-like protease is responsible for the degradation of DNA-PKcs. We have shown that CPP32 cleaves purified DNA-PKcs, producing a fragmentation pattern similar to that observed in cells undergoing apoptosis. This pattern is also observed when DNA-PKcs is incubated with extracts from apoptotic cells. In addition, N-terminal sequencing demonstrates that the site of cleavage, which generates a 150 kDa fragment, is identical in CPP32-digested DNA-PKcs and when extracts from cells undergoing apoptosis are used. A sequence analysis identified the primary cleavage site in DNA-PKcs at DEVD4IN (2709-2713), which is closely related to the DEVD4-G cleavage site in PARP (Lazebnik et al., 1994). This concordance between in vitro and in vivo data supports a role for CPP32 or a related ICElike protease in DNA-PKcs cleavage during apoptosis. It seems unlikely that ICE, TX or Mch2a fulfil this role because none of these enzymes is capable of cleaving purified DNA-PKcs. DNA-PKcs is only the second substrate identified for CPP32 during apoptosis. The other substrate, PARP, which appears not to be directly involved in DNA repair, plays an important role in detecting breaks in DNA and may assist in the maintenance of genomic stability (de Murcia and de Murcia, 1994). On the other hand, DNA-PK binds to free ends in DNA through its DNA-binding component Ku, which leads to the activation
of DNA-PKcs (Dvir et al., 1992; Gottlieb and Jackson, 1993). Because several cell lines lacking components of DNA-PK are deficient in double-strand break repair and are hypersensitive to DNA-damaging agents, this suggests that this complex plays a direct role in the repair of DNA strand breaks (Taccioli et al., 1994; Blunt et al., 1995; Boubnov et al., 1995; Kirchgessner et al., 1995; LeesMiller et al., 1995; Peterson et al., 1995). The degradation of DNA-PKcs, as observed here during apoptosis, would be expected to lead to a reduction in the DNA repair capacity of the cell, favouring the characteristic DNA fragmentation associated with apoptosis. It is also significant that the catalytic subunit alone is degraded; under these conditions the DNA-binding component (Ku) of the complex remains intact. In this context, it is notable that the digestion of nuclei with DNase I, an enzyme implicated in DNA fragmentation in apoptosis (Peitsch et al., 1993), releases the Ku complex which migrates with a subfraction of nucleosomes lacking HI histones (Yaneva and Busch, 1986). These results suggest that Ku is associated with unfolded nucleosomes in DNase I-accessible regions of chromatin and that the binding of Ku to DNA protects against nuclease digestion. It is conceivable that once the catalytic subunit of DNA-PK is degraded, access to nucleases, such as the Ca2+/Mg2+ endonuclease implicated in DNA fragmentation, is increased, which would facilitate DNA degradation during apoptosis. Furthermore, the fact that the pattern of degradation in vitro by CPP32 resembles that observed in cells undergoing apoptosis only when DNA-PKcs is associated with Ku in the presence of DNA, suggests that CPP32 is an attractive candidate for DNAPKcs cleavage in vivo. DNA-PKcs can be added to a short list of key substrates degraded during the process of apoptosis (Kumar and Harvey, 1995; Martin and Green, 1995; Kumar and Lavin, 1996). These include PARP (Kaufmann et al., 1993; Lazebnik et al., 1994), nuclear lamin (Lazebnik et al., 1995; Neamati et al., 1995), the 70 kDa protein component of small nuclear ribonucleoprotein (Casciola-Rosen et al., 1994), fodrin (Martin et al., 1995), actin (Mashima et al., 1995), topoisomerase 1 (Kaufmann, 1989; VoelkelJohnson et al., 1995) and PKC6 (Emoto et al., 1995). The degradation of some of these substrates is compatible with the nuclear disassembly observed in apoptosis. It seems likely that not all of these substrates are cleaved by the same protease. PARP and DNA-PKcs appear to be cleaved by CPP32, but lamin cleavage during apoptosis requires the action of another ICE-like protease (Lazebnik et al., 1995). The inhibition of lamin proteinase prevents nuclear disassembly prior to the packaging of apoptotic bodies but does not interfere with the oligonucleosomal fragmentation of chromatin. These results suggest that lamin degradation is a late step in the apoptotic process. Because DNA-PKcs and PARP are implicated in maintaining genomic integrity and thus in preventing DNA fragmentation, the degradation of these proteins would therefore be expected to precede that of lamin. Indeed, the kinetics of DNA-PKcs and PARP degradation are consistent with this idea. At present, the list of degraded substrates is relatively small but clearly there will be others that are degraded in a manner critical for apoptosis. In the longer term, it is likely that all proteins will be degraded but, as is evident from our observations with Coomassie blue-stained gels,
3243
Q.Song et al.
most proteins remain intact, although most of the cells are undergoing apoptosis. As described by Martin and Green (1995), 'Eventually too many cuts in too many places results in critical change in the cell and it suddenly collapses into apoptotic death'. It remains to be determined to what extent the increasing number of ICE-like proteases isolated are specific in their targeting of individual substrates or how much redundancy is built into the overall process. Since apoptosis is an active process requiring energy input (Schwartzmann and Cidlowski, 1993), the degradation of both DNA-PK and PARP would help to conserve energy for this process. In the absence of a protective mechanism, the fragmentation of DNA, an inherent characteristic of apoptosis, would activate both enzymes which are abundant in the cell (>5 XlO' copies/cell), leading to the consumption of ATP for NAD synthesis for PARP activity, as well as a requirement for ATP by DNA-PK to facilitate the phosphorylation of a variety of physiological substrates (Gottlieb and Jackson, 1993). The cleavage of both of these substrates would be of considerable benefit for apoptosis to proceed in an environment where DNA repair was overwhelmed by the amount of DNA damage. Because PARP knockout mice develop normally, it seems unlikely that the cleavage of this protein is essential for apoptosis (Wang et al., 1995). As with PARP, it is likely that DNA-PKcs is not essential because SCID mice lacking active DNA-PKcs are capable of developing essentially normally. These data suggest that the process of apoptosis probably requires the coordinate degradation of multiple cellular targets by ICE family proteases. The involvement of multiple proteases with different substrate specificities favours the existence of several pathways which may act in a cooperative fashion to lead to apoptosis. However, not all arms of these pathways need be operative in a specific set of circumstances.
Materials and methods Reagents Aprotinin, leupeptin, E-64, pepstatin, TLCK and TPCK were obtained from Boehringer Mannheim. PMSF, iodoacetamide and N-ethylmaleimide were obtained from Sigma and YVAD-CMK from Bachem. Protease inhibitor stock solutions were made up as follows: PMSF, 0.2 M in methanol; aprotinin, 10 mg/ml in water; leupeptin, I mg/ml in water; pepstatin, I mg/ml in methanol; E-64, 10 mg/ml in a mixture of ethanol/water; TLCK, 20 mM in methanol; TPCK, 40 mM in ethanol; iodoacetamide, 100 mM in dimethyl sulfoxide (DMSO); N-ethylmaleimide, 100 mM in DMSO; and YVAD-CMK, 2 mM in DMSO. Purified recombinant ICE was a kind gift from Nancy Thornberry (Rahway, NJ). Apoptosis-inducing agents (etoposide, cisplatin and EGTA) were obtained from Sigma. Tetrandrine is an anti-inflammatory, immunosuppressive compound extracted from the root of the creeper Stephania tetranchra S capable of causing apoptosis in lymphoid cells (Teh et al., 1991). The following antibodies were used for the immunoblotting of DNAPKcs and Ku7O and Ku8O: AbSLY, a polyclonal antibody against amino acid region 1550-1840 of DNA-PKcs; DPK1, a polyclonal antibody against amino acid region 2018-2136; Ab9607, a polyclonal antibody against amino acid region 2790-3065; Ab42-27, a monoclonal antibody that recognizes the kinase (exact sequence undefined) domain; and AbFLA, an antibody against the whole molecule. A mouse polyclonal antibody that detects both Ku7O and Ku8O, as described previously (Chan et al., 1995), was also used.
Cell lines and culture conditions
The Burkitt's lymphoma cell lines BL30A, BL3OK, BM13674, WW2 and BL29, the monocytic cell line U937, the lymphocytic leukemia
3244
Molt-4 cell line and HeLa cells were maintained in RPMI-1640 medium containing 10% heat-inactivated fetal calf serum at 37°C under a humidified atmosphere of 5% CO,. Cells were induced to undergo apoptosis by exposure to etoposide (68 tM). ethanol (5%), cisplatin (20,tg/ml), tetrandrine (10,tg/ml) or 20 Gy of y-rays from a 137cs source (4 Gy/min).
Detection of apoptosis Cell morphology was evaluated using fluorescence microscopy. At the end of each incubation, cells were pelleted at 200 g for 5 min and resuspended in fresh RPMI-1640 medium containing 10% fetal calf serum. The cells were stained with 4',6-diamidin-2-phenylindol-dihydrochloride (I gg/ml) and observed using a Zeiss Axioskop-20 model fluorescence microscope. The extent of apoptosis was determined by the number of cells undergoing micronuclear fragmentation and condensation.
DNA extraction and electrophoresis DNA was extracted from cells according to the procedure described by Miller et al. (1988). Approximately 10 tg of DNA were loaded into each well and electrophoresis was carried out at 10 mA in 1.3% agarose gels with TBE buffer (89 mM Tris-HCI, 89 mM boric acid, 3 mM EDTA, pH 8.0). After electrophoresis, each gel was stained in water containing 2,ug/ml ethidium bromide for 10 min and destained in water for 10 min. The DNA was visualized by UV illumination.
DNA-PK activity assay
Whole-cell extracts were prepared from BL30A cells treated with 68,uM etoposide and assayed with a synthetic peptide (PESQEAFADLWKK) at 0.25 mM, as described previously (Allalunis-Turner et al.. 1995). Each sample was also assayed in the presence of non-substrate peptide EPPLSEQFADLWKK. The incorporation of phosphate into this peptide was
